A solid‐phase transfection platform for arrayed CRISPR screens

Abstract Arrayed CRISPR‐based screens emerge as a powerful alternative to pooled screens making it possible to investigate a wide range of cellular phenotypes that are typically not amenable to pooled screens. Here, we describe a solid‐phase transfection platform that enables CRISPR‐based genetic screens in arrayed format with flexible readouts. We demonstrate efficient gene knockout upon delivery of guide RNAs and Cas9/guide RNA ribonucleoprotein complexes into untransformed and cancer cell lines. In addition, we provide evidence that our platform can be easily adapted to high‐throughput screens and we use this approach to study oncogene addiction in tumor cells. Finally demonstrating that the human primary cells can also be edited using this method, we pave the way for rapid testing of potential targeted therapies.

A Immunoblots showing inducible Cas9 expression in RPE-1 and HEK293T cell lines after 24 h of doxycycline induction (100 ng/ml). B Cell lines stably expressing Cas9-GFP were imaged using transmitted and fluorescent light. Scale bar, 100 lm. C Cell lines expressing inducible Cas9 were stained using anti-Cas9 (green) antibody as well as Phalloidin (red) and Hoechst (blue) to mark actin and DNA, respectively.
Cells were fixed after 48 h of Cas9 induction. Scale bars, 100 lm.
▸ Figure EV2. Characterization of gRNAs used to establish the solid-phase transfection platform.
A Solid-phase transfection of nontargeting (scrambled) or PLK1 targeting siRNA complexes into RPE-1 cells. Cells were fixed after 24, 48, and 72 h and imaged after DNA staining with Hoechst. Green arrowheads show representative cells arrested in prometaphase, and the red arrowheads show representative dead cells due to Plk1 downregulation. Scale bar, 20 lm. B Quantification of experiments in Fig 2A and (A). C Solid-phase transfection of PLK1 targeting gRNAs or RNP complexes into Cas9-expressing RPE-1 or WT RPE-1 cells, respectively. Cells were lysed 24 h posttransfection, and gene editing at the relevant gene loci was assessed by Surveyor assay. Arrowheads indicate the correct size of the digested fragments by the Surveyor nuclease. D Solid-phase transfection of GOLGA2 targeting gRNAs or RNP complexes into Cas9-expressing RPE-1 or WT RPE-1 cells, respectively. Cells were processed and analyzed as in (C). Arrowheads indicate the correct size of the digested fragments by the Surveyor nuclease. E Solid-phase transfection of MKI67 targeting gRNA complexes into Cas9-expressing RPE-1 cells. Cells were fixed after 72 h, stained with Ki67 antibody (red), and analyzed by flow cytometry. GM, geometric mean of the signal intensity. F Cas9-expressing RPE-1 cells were either mock transfected or transfected with nontargeting or PLK1, or POLR2A targeting gRNAs. Five days post-transfection, cell viability was measured by luminescent signal based on ATP production using CellTiter-Glo. The raw values are background subtracted and normalized to the untransfected controls. Results are from three independent experiments containing three technical replicates. In the boxplots, centerlines mark the medians, box limits indicate the 25 th and 75 th percentiles, and whiskers extend to 5 th and 95 th percentiles. P values (scrambled versus Plk1) and (scrambled versus POLR2A) < 0.001. G Solid-phase transfection of POLR2A targeting gRNAs or RNP complexes into Cas9-expressing RPE-1 or WT RPE-1 cells, respectively. Cells were processed and analyzed as in (C). Arrowheads indicate the correct size of the digested fragments by the Surveyor nuclease.   A Solid-phase transfection of mock and scrambled gRNA in NCI-H358 and NCI-N87, RPE-1, and HEK293T cells. Transmission images of the cell lines were acquired 72 h post-transfection. Note that there are no visible fitness defects upon transfection. Scale bar, 400 lm. B NCI-H358 and NCI-N87, RPE-1 and HEK293T cells were transfected with scrambled or POLR2A targeting gRNAs on plates that were stored for 3 weeks at room temperature. Five days post-transfection, cell viability was measured by CellTiter-Glo. Boxplots represent values from three independent experiments containing three technical replicates. Centerlines mark the medians, box limits indicate the 25 th and 75 th percentiles, and whiskers extend to 5 th and 95 th percentiles. C NCI-H358 and NCI-N87, RPE-1 and HEK293T cells were transfected with indicated control gRNAs. Five days post-transfection, cell viability was measured by CellTiter-Glo. Boxplots represent values from three independent experiments containing three technical replicates. In the boxplots, centerlines mark the medians, box limits indicate the 25 th and 75 th percentiles, and whiskers extend to 5 th and 95 th percentiles. D Cell viability measurements after mock transfection in a panel of cell lines described in Fig 2I. Boxplots represent values from three independent experiments.
Centerlines mark the medians, box limits indicate the 25 th and 75 th percentiles, and whiskers extend to 5 th and 95 th percentiles.   A Calculation of LOD scores for effect on viability of guide RNA transfection: (i) Normalized values of viability were obtained by ratio of each guide to scrambled control transfection. (ii) Normal distribution of probabilities for each guide was calculated using cumulative normal distribution function using normalized values and mean values of controls. LOD scores were calculated for each guide using values generated in (iii) using the formula shown. B An example of LOD score and viability graph. Density graphs show positive and negative control guides and their normal distribution. Guide A targets a gene causes strong loss of viability with a high LOD score, whereas guide B shows no loss of viability, with a low LOD score. C Controls for normality. Distribution and qqplots of control guides used in our study. Distributions of the control guides were tested by a Kolmogorov-Smirnov test.
We have not observed any significant differences to normally distributed samples. D Prediction of proteins that can be targeted by arrayed CRISPR/Cas9-based screens. Using the dataset from McShane et al, we considered 1-state degradation model for exponentially degraded proteins (ED) and 2-state degradation model for non-exponentially degraded (NED) proteins. The densities of all the proteins are plotted and classified as "targetable" and "hard to target" in CRISPR/Cas9-based arrayed screens based on a threshold of 120-h half-life.